Highly Available MySQL Server

Standard

Previously, I have discussed how to setup a highly available block device and also a highly available file system.  In this article, I further demonstrate how to setup a highly MySQL database service.

Installing the Packages

As usual, one will need to install the package on two hosts and it can be easily done by:

# pkg install mysql57-server

I know there are alternatives.  Forgive my laziness.

Running for the First Time

We are starting MySQL once so that it generates the file structures.  Try to login, and then Ctrl-D to exit.

# service mysql-server onestart
# cat /root/.mysql_secret
# mysql -u root -p
Password: **********
root@localhost [(none)]> ^D
# service mysql-server onestop

It is then discovered (through educated guess) some directories are created in the “/var/db” directory, namely “mysql”, “mysql_tmpdir”, and “mysql_secure”.  Suppose you already have the “/db” mounted (as in the previous article), move them there and make the replacement symbolic links.

# mv /var/db/mysql /db
# mv /var/db/mysql_tmpdir /db
# mv /var/db/mysql_secure /db
# ln -s /db/mysql /var/db/
# ln -s /db/mysql_tmpdir /var/db/
# ln -s /db/mysql_secure /var/db/
# ls -ld /var/db/mysql*
lrwxr-xr-x  1 root  wheel   9 Apr 19 20:37 /var/db/mysql -> /db/mysql
lrwxr-xr-x  1 root  wheel  16 Apr 19 20:37 /var/db/mysql_secure -> /db/mysql_secure
lrwxr-xr-x  1 root  wheel  16 Apr 19 20:37 /var/db/mysql_tmpdir -> /db/mysql_tmpdir

Some would question why not change the configuration for the new paths.  I find it mostly a matter of taste.  If you want to make lives easier for those who have recited the default paths, do make the symbolic links.

Configurations

You will want to modify the configuration file “/usr/local/etc/mysql/my.cnf”.  For your reference, there is a sample file “my.cnf.sample”.  At minimum, you will need to modify the bind address (default 127.0.0.1) so that the service is available not just locally, but to the other computers in the same intranet.

The Script

The script for starting and stopping the MySQL server is simpler than the NFS one and are as follows.  Like last time, automatic switching is skipped due to my conflict of interest.  You will need a mechanism to call “start” and “stop” properly.

#!/bin/sh -x

start() {
 ifconfig vtnet1 add 10.65.10.14/24
 hastctl role primary db_block
 while [ ! -e /dev/hast/db_block ]
 do
 sleep 1
 done
 fsck -t ufs /dev/hast/db_block
 mount /dev/hast/db_block /db
 service mysql-server onestart
}

stop() {
 service mysql-server onestop
 umount /db
 hastctl role secondary db_block
 ifconfig vtnet1 delete 10.65.10.14
}

status() {
 ifconfig vtnet1 | grep 10.65.10.14 && \
 service mysql-server onestatus && \
 ls /dev/hast/db_block
}

residue() {
 ifconfig vtnet1 | grep 10.65.10.14 || \
 service mysql-server onestatus || \
 mount | grep /db || \
 ls /dev/hast/db_block
}

clean() {
 residue
 if [ $? -ne 0 ]
 then
 exit 0
 fi
 exit 1
}

if [ "$1" == "start" ]
then
  start
elif [ "$1" == "stop" ]
then
  stop
elif [ "$1" == "status" ]
then
  status
elif [ "$1" == "clean" ]
then
  clean
fi

Troubleshoot

If there are any issues MySQL fails to start, you can verify its absence with the command “service mysql-server onestatus”.  There are also log files located in the MySQL data directory; in our context, it is “/db/mysql/<hostname>.err”.  Please note the end of the log is most likely a graceful shutdown.  You will need to scroll upwards for the actual reason why the startup failed.

Advertisements

System Performance with FreeBSD (Minecraft Server as Example)

Standard

Quite some time ago, we discussed how to get compile Minecraft and get it running on FreeBSD.  In this article, we take the server as an example how we can monitor system performance.

Minecraft Memory Usage

Minecraft is a Java program.  Java programs generally consume more memory than the counterparts made of unmanaged languages.  Thankfully, Java programs run inside their own sandboxes and have memory usage allocated and constrained.  In the previous article, we defined memory to be 1024 megabytes and expands up to 1024 megabytes only:

java -Xmx1024M -Xms1024M -jar spigot*.jar

It is important not to have Minecraft overrun the system memory.  As folklore, a Java program running on a dedicated computer should not be higher than 60% of total memory.  For example, my virtual machine has 2048 megabytes of memory and 60% of it is about 1200 megabytes.  I deducted myself further for 200 megabytes as safety margin.

General Process Monitoring

FreeBSD provides top(1) utility to check for system saturation and utilisation.  Generally speaking, a system is considered saturated if the number of threads ready to run is higher than the number of processor cores, and considered fully utilised if the utilisation numbers are near 100%.

螢幕快照 2017-06-06 下午10.57.56.png

In the top part, there are three numbers in decimal point, representing the load averages in 1, 5, and 15 minutes.  This is calculated by average number of threads ready to run and is regarded as the saturation of the processors.  In the third and forth row, the overall processor and memory utilisation ratios / rates are shown.

In the picture, the system not yet saturated as there is about 0.6 threads to run per second.  It is under some computation load that around 20% processing power and 80% memory are utilised.  This is a healthy situation that the system is being used, with some slacks to handle possible usage spikes.

In the bottom part, we have detailed breakdown per process.  The fields “TIME” and “WCPU” represents the total time and current portion of processor each process is using. Actual amount of memory usage are represented in the “RES” column.

In the picture, we see the Java process using 44% of current processing power and it has accumulated 295 hours.  Although the Java process is instructed to use only 1024 megabytes of memory, eventually it consumes 1368 megabytes.  This echoes why the folklores recommend only specify 60% of system memory to a Java virtual machine.

System Input / Output Monitoring

FreeBSD provides systat(1) utility for system usage statistics.  Since the general process monitoring can be done by the top(1) command above, it is mostly useful for detailed input and output monitoring.  By default, it shows a “pig” screen that refreshes every 10 seconds.  You can specify what you want to see by specifying in commands, for example, to see network interface usage every second:

systat -ifstat 1

螢幕快照 2017-06-06 下午11.02.03

The screen works like vi(1).  You can press the colon sign (:) and to enter a command.  The first command you want to try is “help”.  It immediate tells you the list of available pages you can switch to.  You can then use the colon sign again and enter the page you want.

I would tell you vmstat is my favourite page.  I learned it on the very first days I was taught FreeBSD.  It shows a lot of comprehensive information from system utilisation to interrupts and disk accesses.

systat -vmstat 1

螢幕快照 2017-06-06 下午10.59.11On the top, we see the system saturation and utilisation as usual.  Immediately under it, we see the detailed breakdown system events like context switches (Csw), traps (Trp), system calls (Sys), etc.  The system is now having thousands of context switch per second.  It would be no good if it were a system for high performance computing, but for our context of gaming with network messages, it is absolutely normal.

Further below, we see an ASCII art of system utilisation break down into system (=), interrupts (+), user space applications (>) and niced user applications (-).  The system is using mild amount of processing power and most of it are for the user space, which is a good thing.

The second bottom section shows the name caches of the virtual file system.  For a gaming system that uses few files, you can expect the hit rate be near 100%.  Otherwise you may want to pay more attention to the name caches.  In the picture, we see the system rarely searches for a file and those requests are handled by the cache perfectly.

The bottom section shows the disk utilisation.  Utilisation and bandwidth of each of the virtual disks are shown.  In the picture, we see the system rarely access the files.

The right most column shows the interrupt statistics.  The “timer” interrupts happens almost all the time in order to hint the operating system to context switch and update system clocks.  It used to be 100 or 1000 per processor core; thankfully, with the advent of more advanced system clocks, systems no longer need to tick as frequent as before.  In the picture, the network card (virtio-network) requires quite some interrupts handling.  As long as the network card interrupts does not go to numbers like 50000 or 100000, they are most likely normal.

The list of pages available in the tool, as of today, are:

  • pigs: shows the processes which consume the most processing power
  • vmstat: (as discussed above)
  • swap: shows the system swap situation
  • zarc: shows the ZFS adaptive read cache situation
  • iostat: raw disk input and output statistics
  • netstat: network socket statistics, such as buffered bytes for each of the connections
  • sctp: stream control transport protocol statistics
  • tcp: transport layer protocol statistics
  • ip: internet protocol statistics
  • ip6: internet protocol statistics for IPv6
  • icmp: internet control message statistics such as ping, etc
  • icmp6: internet control message statistics for IPv6
  • ifstat: raw network interface utilisation statistics

Conclusion

In this article, we go through some performance monitoring tools that come with FreeBSD.  The general process information can be listed by the top(1) command, where you can understand the system saturation and utilisation, and also list the resource consuming processes.  More detailed resource utilisations like network, disks, hardware interrupts can be found in the systat(1) command.  If in doubt, the “vmstat” page can be a great starting point to look for congested system resources.

Highly Available Network File System

Standard

In the previous article, we discussed the way to create a highly available block device by replication.  We continue and attempt making a network file system (NFS) on top of it.  We first discuss the procedures to start and stop the service.  Then we have the script…  Some parts are deliberately missing due to my conflict of interest.

NFS Configuration

Since it is not our goal here, we only do minimal NFS configuration in this example.  In short, the export(5) file “/etc/exports” is being modified like as follows.  This implies the directory “/nfs” is shared with the given two IP subnets.

/nfs -network=10.65.10.0/24
/nfs -network=127.0.0.0/24

Unlike previous setting, we do not use the “/etc/rc.conf” file to start the service.  This is because we like to control when a service is started, instead of blindly just after boot.  In FreeBSD, services can be started with the “onestart” command.

Firewall Configuration

Configuring NFS for a tight firewall is tricky, because it uses random ports.  For convenience, a simple IP address-based whitelist can be implemented.  (It is possible to fix the ports if you are willing to take extra steps).  In this example, we have the server IP 10.65.10.13 (see later), and the client IP 10.65.10.21.  If you simply do not have a firewall, skip this part.  On the server side, the PF can be configured with:

pass in quick on vtnet1 from 10.65.10.21 to 10.65.10.13 keep state

On the client side, the PF can be configured with:

pass in quick on vtnet1 from 10.65.10.13 to 10.65.10.21 keep state

Starting the Service

When we start the service, we want the following to happen:

  1. Acquire the IP address, say 10.65.10.13, regardless which machine it is running.
  2. Activate the HAST subsystem so to become the primary role.
  3. Wait for the HAST device to be available.  If the device is in secondary role, the device file in “/dev/hast” will not appear so we can go to sleep a while.
  4. Run the file system check just in case the file system was corrupted in the last unmount.
  5. Mount the file system for use (in this example, “/nfs”)
  6. Start the NFS-related services in order: the remote procedural call binding daemon, the mount daemon, the network file system daemon, the statistic daemon, and the lock daemon.
  7. Once the step 5 completes, the service is available to the given clients as instructed to the NFS and allowed by the firewall.

For the inpatient, one can jump to the second last section for the actual source code.

Stopping the Service

Stopping the service is the reverse of starting, except some steps can be less serious.

  1. Stop the NFS-related services in order: the lock daemon, the statistic daemon, the network file system daemon, the mount daemon, and finally the remote procedural call binding daemon.
  2. Unmount the file system.
  3. Make the HAST device in secondary role.
  4. Release the iP address so neighbours can reuse.

Also, one can jump to the second last section for the actual source code.

Service Check Script

There are two types of checking.  The first one ensures all the components (like the IP address, mount point service, etc) are present and valid.  The procedure returns success (zero) only when the components are all turned on.  Whenever a component is missing, it will be reported as a failure (non-zero return code).

The second one ensures all the components are simply turned off, so that the service can be started on elsewhere.  The procedure returns success (zero) only when all the components are turned off.  Whenever a component is present, it will be reported as a failure (non-zero return code).

What is Missing

Once we master how to start and stop the service on one node, we need the mechanism to automatically start and stop the service as appropriate.  In particular, it is utmost important not to run the service concurrently on two hosts, as this may damage the file system and confuse the TCP/IP network.  This part should be done out of the routine script.

The Script

Finally, the script is as follows…

#!/bin/sh -x

start() {
  ifconfig vtnet1 add 10.65.10.13/24
  hastctl role primary nfs_block
  while [ ! -e /dev/hast/nfs_block ]
  do
    sleep 1
  done
  fsck -t ufs /dev/hast/nfs_block
  mount /dev/hast/nfs_block /nfs
  service rpcbind onestart
  service mountd onestart
  service nfsd onestart
  service statd onestart
  service lockd onestart
}

stop() {
  service lockd onestop
  service statd onestop
  service nfsd onestop
  service mountd onestop
  service rpcbind onestop
  umount /nfs
  hastctl role secondary nfs_block
  ifconfig vtnet1 delete 10.65.10.13
}

status() {
  ifconfig vtnet1 | grep 10.65.10.13 && \
  service rpcbind onestatus && \
  showmount -e | grep /nfs && \
  mount | grep /nfs && \
  ls /dev/hast/nfs_block
}

residue() {
  ifconfig vtnet1 | grep 10.65.10.13 || \
  (service rpcbind onestatus && showmount -e | grep /nfs) || \
  mount | grep /nfs || \
  ls /dev/hast/nfs_block
}

clean() {
  residue
  if [ $? -ne 0 ]
  then
    exit 0
  fi
  exit 1
}

if [ "$1" == "start" ]
then
  start
elif [ "$1" == "stop" ]
then
  stop
elif [ "$1" == "status" ]
then
  status
elif [ "$1" == "clean" ]
then
  clean
fi

Testing

To test, fine the designated computer and mount the file system.  Assume the file system has been running on the host “store1”, make a manual failover to see…  The file client does not need to explicitly remount the file system; it cab be remounted automatically.

client# mount 10.65.10.13:/nfs /mnt
client# ls /mnt
.snap
client# touch /mnt/helloworld
store1# ./nfs_service.sh stop
store2# ./nfs_service.sh start
client# ls /mnt
.snap helloworld

Highly Available Storage Target on FreeBSD

Standard

To achieve highly available service, it is vital to have the latest data available for restarting the service elsewhere.  In enterprise environments, multipath SAS drives allows each drive to be accessible from multiple hosts.  What about the rest of us, living in the cloud of / or inexpensive SATA drives?  Highly Available Storage Target (HAST) is the answer.  It relatively a new feature in FreeBSD 8.1.  It is useful to keep two copies of drive content on two loosely coupled computers.

In this article, I demonstrate how HAST can be setup, without bothering the actual failover logic.  I assume the two virtual machines are prepared from scratch with some unallocated space.  Alternatively, you can prepare a virtual machine with multiple drives (which is not quite feasible in my setting).

Preparing the Partitions

First, examine the drives.  Here we have 18 (no, indeed 17.9 something) gigabytes of space available.

# gpart show
=>      40  52428720 vtbd0 GPT (25G)
        40       984       - free - (492K)
      1024      1024     1 freebsd-boot (512K)
      2048   4192256     2 freebsd-swap (2.0G)
   4194304  10485760     3 freebsd-ufs (5.0G)
  14680064  37748696       - free - (18G)

Then, we can add two more partitions.  Since we have two hosts, we set two partitions so that each host can run the service on one partition.  This is so-called active-active setup.  I dedicate one for NFS and one for database.  Once you finish working on one virtual machine, do not forget performing the same on another machine.

# gpart add -t freebsd-ufs -l nfs_block -a 1M -s 8960M /dev/vtbd0
vtbd0p4 added.
# gpart add -t freebsd-ufs -l db_block -a 1M -s 8960M /dev/vtbd0
vtbd0p5 added.

This is the result.  Two block devices are created and made available inside /dev/gpt with their appropriate labels.

# gpart show
=>      40  52428720 vtbd0 GPT (25G)
        40       984       - free - (492K)
      1024      1024     1 freebsd-boot (512K)
      2048   4192256     2 freebsd-swap (2.0G)
   4194304  10485760     3 freebsd-ufs (5.0G)
  14680064  18350080     4 freebsd-ufs (8.8G)
  33030144  18350080     5 freebsd-ufs (8.8G)
  51380224   1048536       - free -  (512M)
# ls /dev/gpt
db_block nfs_block

HAST Daemon Setup

Here is an sample of defining the HAST configuration, hast.conf(5).  In short, there are two hosts and two resource items.  The host “store1” has its remote partner “store2” and vice versa.  Since we use the short host names “store1” and “store2”, do not forget to update the host(5) file to make them resolvable.  Remember to repeat these another machine.  Thankfully, the HAST configuration need not to be customised for each member host.

# sysrc hastd_enable="YES"
# cat > /etc/hast.conf << EOF
resource nfs_block {
  on store1 {
    local /dev/gpt/nfs_block
    remote store2
  }
  on store2 {
    local /dev/gpt/nfs_block
    remote store1
  }
}
resource db_block {
  on store1 {
    local /dev/gpt/db_block
    remote store2
  }
  on store2 {
    local /dev/gpt/db_block
    remote store1
  }
}
EOF
# service hastd start

Firewall Rules

If you are having a firewall, remember to open the port number 8457 opened.  For example, in PF, add these three lines to the two hosts.  Remember the replace the IP addresses as appropriate.

geompeers="{10.65.10.11,10.65.10.12}"
geomports="{8457}"
pass in quick inet proto tcp from $geompeers to any port $geomports keep state

HAST Daemon Status

Once the HAST daemon is started, the status of the blocks can be checked.  Since we have defined two resource items, there are two HAST device status reported.  For example, in the host “store1”, it says there are two components for each resource item: one block device, and one remote host.  At first, the resource items are in “initialisation” state:

store1# hastctl status
Name      Status   Role      Components
nfs_block -        init      /dev/gpt/nfs_block store2
db_block  -        init      /dev/gpt/db_block store2

To turn on the device for operation, use the “role” subcommand on “store1”

store1# hastctl role primary db_block
store1# hastctl status
Name      Status   Role      Components
nfs_block -        init      /dev/gpt/nfs_block store2
db_block  degraded primary   /dev/gpt/db_block store2

Similarly, use the “role” command on “store2”, but this time we set it secondary:

store2# hastctl role secondary db_block
store2# hastctl status
Name      Status   Role      Components
nfs_block -        init      /dev/gpt/nfs_block store1
db_block  -        secondary /dev/gpt/db_block store1

When the synchronisation completes, the status is marked complete:

store2# hastctl status
Name      Status   Role      Components
nfs_block -        init      /dev/gpt/nfs_block store1
db_block  complete secondary /dev/gpt/db_block store1

Formatting the Block Devices

We can format the block device like usual, just that we should format the device under the device directory “/dev/hast” instead of “/dev/gpt”.  Since currently the “db_block” is active on the host “store1”, it has to be executed over there.

store1# newfs -J /dev/hast/db_block
/dev/hast/db_block: 8960.0MB (18350064 sectors) block size 32768, fragment size 4096
using 15 cylinder groups of 626.09MB, 20035 blks, 80256 inodes.
super-block backups (for fsck_ffs -b #) at:
192, 1282432, 2564672, 3846912, 5129152, 6411392, 7693632, 8975872, 10258112, 11540352, 12822592, 14104832, 15387072, 16669312, 17951552

One thing to note is, the raw device was 18350080 sectors.  When HAST takes it for service, there is only 18350064 blocks left for payload.  Next, we can mount the file system.  We do not use the fstab(5) like before, because they do not need to execute every time boot.

store1# mkdir /db
store1# mount /dev/hast/db_block /db

Switching Over

In order to switch over, the procedure is as follows.

store1# umount /db
store1# hastctl role secondary db_block
store2# hastctl role primary db_block
store2# mkdir /db
store2# mount /dev/hast/db_block /db

What if, in an error, the backend device in “/dev/gpt” is being used for mounting?  It will say the following.  The chance to go wrong is really not heavy.

store2# mount /dev/gpt/db_block /db
/dev/gpt/db_block: Operation not permitted

For automatic switching, it will be discussed in a separated article.

Troubleshooting

Nothing can go really wrong without a serious application on top.  Nevertheless, the following message troubled me for a few hours.

We act as primary for the resource and not as secondary as secondary“: check the HAST configuration.  Likely a host is configured to have itself as the remote partner.

To be Continued

In the upcoming articles, I will cover how to make use of this highly-available block storage for shared file system, database, and web servers.

Installing FreeBSD from Scratch and Reinstalling the Boot Loader

Standard

There are cases the default image does not suit for one.  In this exercise, I practice installing FreeBSD version 11 from scratch.  I go beyond the standard procedure by partitioning the drive manually with commands. This is to leave space I can create partitions purely for payload later.   (If you just want to go automatic, you can refer to the FreeBSD handbook.)

Some errors take place so I get to correct the boot loader manually.  If you have tried fixing the boot loader of some other “freedom” operating system, you will appreciate how easy it is!

Inserting the Disc and Boot

Instead of selecting the default boot image, we pick an installation disc.  In Vultr, There are two ways.  The first way is to let the system download the installation disc.  For example, you find a link for the FreeBSD installation disc, copy the URL, and pass it to the interface.  The second way is to reuse the existing library of installation discs.

It takes quite some time for the system to boot.  Depending whether you are lucky or not, you may or may not see the beastie welcome screen.  This is so-called the boot loader, or simply the loader, with just a few tens of kilobytes.

Screen Shot 2017-04-13 at 9.33.34 pm

Inside the Installer

The system boots and the installer (precisely, “bsdinstall”) automatically executes.  From now on, there are a few keystrokes you need to know.  The action buttons, quoted in brackets, can be selected with left and right arrow keys.  To toggle the action button, press enter key.  The items above the action buttons are selected with up and down.  To toggle the item on or off, press spacebar.  At any one time, an action button and a selectable item are highlighted.  When there are multiple fields, press the tab, not enter, to jump between.

Question 1 – mode selection: In the screen below, you can press enter to run the installer.  You can alternatively press right arrow to select the shell, then enter to run the shell.  Here we select “install” directly.

Screen Shot 2017-04-13 at 9.34.07 pm

Question 2 – keymap: If you want to select an alternative keymap, use up and down arrow keys, and press spacebar to select.  Then, press enter to confirm.

Screen Shot 2017-04-13 at 9.34.17 pm

Question 3 – hostname: You are going to enter a hostname.  If you are creating a machine to be cloned, you can pick a generic name.

Question 4 – distributions: You are asked what distribution components to select.  Usually I just pick “lib32” only.  By default, they propose installing “ports”, I deselect it (with spacebar) most of the time.  The updated ports can be downloaded by “postsnap” command later.

Partitioning and Formatting the Drive

Question 5 – partition method: You are given several ways to partition, the “auto” one are the most easy but they may generate something you do not like.  The “manual” shows a dialog where you can create the partitions yourself, but not control the partition alignments.  So let us select “shell”.

Screen Shot 2017-04-13 at 9.35.40 pm.png

Question 6 – partition: You are given a shell and instructed to type in commands, edit a file, and mount the effective file system.  Use the following commands to partition the only virtual hard drive, “vtbd0”, and then install the bootloader.

Screen Shot 2017-04-13 at 9.35.50 pm

# gpart show
# gpart create -s gpt /dev/vtbd0
vtbd0 created
# gpart show
=>      40  52428720 vtbd0 GPT (25G)
        40  52428720       - free - (25G)

# gpart add -t freebsd-boot -a 512K -s 512K /dev/vtbd0
vtbd0p1 added
# gpart add -t freebsd-swap -a 1M -s 2047M /dev/vtbd0
vtbd0p2 added
# gpart add -t freebsd-ufs -a 1M -s 5120M /dev/vtbd0
vtbd0p3 added
# gpart show
=>      40  52428720 vtbd0 GPT (25G)
        40       984       - free - (492K)
      1024      1024     1 freebsd-boot (512K)
      2048   4192256     2 freebsd-swap (2.0G)
   4194304  10485760     3 freebsd-ufs (5.0G)
  14680064  37748696       - free - (18.0G)
# gpart bootcode -b /boot/pmbr -p /boot/gptboot -i 1 /dev/vtbd0
bootcode written to /dev/vtbd0

Previous step, we partition the drive into three, a boot partition, a swap partition, and a unix file system partition.  We install the GPT boot loader into the boot partition.  Then, format the last partition, define the file system table as previously instructed, then we are done.  The installer starts installation without a question asked.

# newfs -U /dev/vtbd0p3
(message truncated)

# mount /dev/vtbd0p3 /mnt
# cat >> /tmp/bsdinstall_etc/fstab << EOF
/dev/vtbd0p2 none swap sw 0 0
/dev/vtbd0p3 /    ufs  rw 1 1
EOF

# exit

Screen Shot 2017-04-13 at 9.59.14 pm

Final Touches to the Installation

Question 7 – root password: Pick and enter a password carefully, twice.

Question 8 – network configuration: You are asked what network devices you like to configure.  Select the only virtual network device, “vtnet0”.  Enable IPv4 and DHCP.  Disable IPv6 (unless you know why not).

Question 9 – name resolver configuration: Simply press “ok” for the DNS configuration.  The DNS server setting will be overridden soon.

Question 10 – time zone selection: Select the continent you are in, and then the city.  You are then asked if the abbreviation is appropriate, and confirm the system date and time.

Question 11 – services: I would select “local_unbound”, “sshd”, and “ntpd”.

Screen Shot 2017-04-13 at 10.01.51 pm

Question 12 – security: Since version 11, the FreeBSD installer asks if the user wants any additional security measures.  I think most of them can be enabled, except the debugging.  (This is because I do debug programs.)

Screen Shot 2017-04-13 at 10.03.21 pm

Question 13 – additional users: This is up to you.  I prefer customisation before user creation.

Question 14 – final configuration: Just skip…

Question 15 – final modification: Just skip…

Question 16 – what next: Instead of rebooting, I prefer going to the live CD mode, login and “poweroff”.

Remaining Activities

Take a snapshot before booting the system again.  On the first system boot, the SSH generates its identities.  If you want multiple hosts having their distinct identities, taking the snapshot before the first boot is the laziest and the most correct way.

Last but not least, remove the virtual optical drive image.  Then you are good to boot from the virtual hard drive.

Troubleshooting and Fixing the Boot Loader

Missing boot loader: When generating the screenshots, I forgot to install the boot code.  The boot screen looks like this and is stuck.  This is a sign of missing the boot loader.  I booted with the installation disc again, then choose shell mode, and finally rerun the “gpart bootcode” command.

Screen Shot 2017-04-13 at 10.05.10 pm

# gpart show
=>      40  52428720 vtbd0 GPT (25G)
        40       984       - free - (492K)
      1024      1024     1 freebsd-boot (512K)
      2048   4192256     2 freebsd-swap (2.0G)
   4194304  10485760     3 freebsd-ufs (5.0G)
  14680064  37748696       - free - (18G)
# gpart bootcode -b /boot/pmbr -p /boot/gptboot -i 1 /dev/vtbd0
bootcode written to /dev/vtbd0

Damaged file system table: On the next boot attempt, I drop into single user mode because of bad file system table.  This was because I wrote “rw” instead of “sw” for the swap.  I then corrected the “/etc/fstab” with an editor.  Then I “exit” to continue the boot.

Screen Shot 2017-04-13 at 10.11.27 pm.png

Security Settings

For you reference, the security options I made in installation turns out to be the following.  So they can be incorporated in other installation tools, without actually running the “bsdinstall”.

/etc/rc.conf

clear_tmp_enable="YES"
syslogd_flags="-ss"
local_unbound_enable="YES"

/etc/sysctl.conf

security.bsd.see_other_uids=0
security.bsd.see_other_gids=0
security.bsd.unprivileged_read_msgbuf=0
security.bsd.stack_guard_page = 1

/etc/resolv.conf

nameserver 127.0.0.1
options edns0

To be Continued

In the upcoming articles, I will use the snapshots created here to build a highly available block device, and then highly available file systems and database systems.

Basic C TCP/IP Programming

Standard

In this article, I share how to have some TCP/IP programming with the C language.  I am using FreeBSD because it is my most familiar platform but it does not prevent you trying the source code elsewhere.  There are of course thousands of examples online.  To show that I am different, I will present my code in an unorthodox but effective way.  (I am a protestant, if you force me to answer.  🙂

I am using the word TCP/IP in the title, in case I want to have things like libibverbs in another time.  In this article, when I use the word “network”, I refer to TCP/IP.

Header Files

Some header files are required, such as standard system call headers, data type definitions, string operations, etc.

#include <stdio.h>
#include <unistd.h>
#include <sys/types.h>
#include <sys/socket.h>
#include <sys/wait.h>
#include <netdb.h>
#include <string.h>

System Calls and File Descriptors

In Unix system, a lot of things are presented as file descriptors and these include the network sockets we discuss here.  There are various system calls to get file descriptors of different types, such as open(2) for a named file, pipe(2) for inter-process communications, and socket(2) for a network socket.  Even so, once a file descriptor is prepared, the operations are similar: read(2) for withdrawing message, write(2) for depositing message, and close(2) for finishing after use.

Things are easy to be said than to be done.  To make a network connection, there are quite some steps that must go through.  Here I split into two roles, a server that receives connection, and a client that initiates connection.

Server:

  1. socket(2) to create a network socket
  2. bind(2) to attach to a particular defined network port
  3. listen(2) to create a connection queue
  4. accept(2) to accept a connection request
  5. read(2) or write(2) to communicate
  6. close(2) to finish

Client:

  1. socket(2) to create a network socket
  2. connect(2) to connect to a particular address
  3. read(2) or write(2) to communicate
  4. close(2) to finish

Obtaining Socket Address

In order to bind or connect, one needs to provide a socket address.  It is a structure with complex data structure.  Some writers would tell readers to fill in the structures one by one, tell them to be caution about the endiness, etc.  I am lazy and just use the getaddrinfo(3) function.  It is also the most reliable method if you want to handle different network types.  It takes the hostname, port number, and optionally hints as input.  It generates a data structure and return the pointer through a pointer of pointer.  The most difficult part is filling in the hints, but these can be blindly copied.  For example, to get a TCP over IPv4, we request SOCK_STREAM and AF_INET like…

memset(&hint, 0, sizeof(hint));
hint.ai_family = AF_INET;
hint.ai_socktype = SOCK_STREAM;
hint.ai_protocol = 0;
getaddrinfo(host, port, &hint, &info);

Given there are not errors, the info structure now contains the answer.  “info->ai_addr” points to the required sockaddr structure, and “info->ai_addrlen” points to the length of the answer.  Just in case you did not define the numbers, also helps by filling in the “info->ai_family,” “info->ai_socktype”, and “info->ai_protocol” according to the hints.  These are useful in creating the first socket.

Please note the address info can be a linked list when there are multiple options for the connection.  To be robust, one may want to try out all the connections.  For demonstration, trying one is enough.  After use, it can be cleanup with the corresponding freeaddrinfo(3) function.

Error Handling

Most system calls come with exceptional situations, like when a file is not found, or a network socket is not connectable.  It is a good practice to check the return value of each system call.  The code becomes very messy and this is when people yell for the “exception” feature of a language — put the error handling code out of the normal execution!  Hold on, checking return value can be trivial with C macro functions.  I learned this trick from a famous book but I do not recall the name.

#define pt {fprintf(stderr, "%s:%d: ", __FILE__, __LINE__); perror("");}
#define ez(x) {if ((x) != 0) {pt; goto error;}}
#define ep(x) {if ((x) <= 0) {pt; goto error;}}
#define ezp(x) {if ((x) < 0) {pt; goto error;}}

Whenever I expect it to return zero, I use ez (expect zero); also ep for expecting positive, and zp for expecting natural numbers.  With these, the error checking can be much easier, for example,

if (listen(fd, 1) != 0) {
        fprintf(stderr, "%s:%d: ", __FILE__, __LINE__);
        perror("");
        goto error;
}

can be replaced as

ez(listen(fd, 1));

At the end of the function, I just define a label to catch these errors and perform cleanup.  This is now even simpler than handling exceptions.  The actual cleanup code will be shown later.

The Server Code

The server first prepares the hints and obtains the socket address accordingly.  Then it calls the socket, bind, listen, accept, write system call accordingly.  There are two file descriptors, one for binding to the particular port, and one (or more) for communicating with the clients.  After use, it cleans up cautiously.  If there is ever error, the flow jumps to “error” immediately and the variable is updated accordingly.

int server(const char* host, const char* port)
{
        int fd = -1;
        int fd2 = -1;
        int error = 0;
        struct addrinfo* info = 0;
        struct addrinfo  hint;
        char message[16] = "hello world";
        memset(&hint, 0, sizeof(hint));
        hint.ai_family = AF_INET;
        hint.ai_socktype = SOCK_STREAM;
        hint.ai_protocol = 0;
        ez(getaddrinfo(host, port, &hint, &info));

        ezp(fd = socket(info->ai_family, info->ai_socktype, info->ai_protocol));
        ez(bind(fd, info->ai_addr, info->ai_addrlen));
        ez(listen(fd, 1));
        ezp(fd2 = accept(fd, 0, 0));
        ep(write(fd2, message, sizeof(message)));

cleanup:
        if (info != 0) freeaddrinfo(info);
        if (fd2 != -1) close(fd2);
        if (fd != -1) close(fd);
        return error;

error:
        error = 1;
        goto cleanup;
}

The Client Code

The client code is similar to the server, except it only has one file descriptor and it connects rather than bind or connect.  After reading the message from the server, it prints it out and finish.

int server(const char* host, const char* port)
{
        int fd = -1;
        int fd2 = -1;
        int error = 0;
        struct addrinfo* info = 0;
        struct addrinfo  hint;
        char message[16] = "";

        memset(&hint, 0, sizeof(hint));
        hint.ai_family = AF_INET;
        hint.ai_socktype = SOCK_STREAM;
        hint.ai_protocol = 0;
        ez(getaddrinfo(host, port, &hint, &info));
        ezp(fd = socket(info->ai_family, info->ai_socktype, info->ai_protocol));
        ez(connect(fd, info->ai_addr, info->ai_addrlen));
        ep(read(fd, message, sizeof(message)));
        printf("client received: %s\n", message);

cleanup:
        if (info != 0) freeaddrinfo(info);
        if (fd2 != -1) close(fd2);
        if (fd != -1) close(fd);
        return error;

error:
        error = 1;
        goto cleanup;
}

Putting Them Together

Finally, we need a main function to run a server and a client about the same time.  Here we use fork(2) and wait(3) calls.  The client is delayed 1 second to ensure the server has got ready before a connection is established.  The code pasting will be left as an exercise.  In short…

  1. The header files
  2. The error handling macros
  3. The server code
  4. The client code
  5. The main function

The main function is as follows.  After fork, there are two processes for the two roles.  The server is started to accept connections from port 8080 of any given IP addresses.  The client is targeting localhost port 8080.

int main(int argc, char** argv)
{
        int status = 0;
        pid_t pid = fork();

        // Forked as a parent
        if (pid > 0) {
                server(0, "8080");
                waitpid(pid, &status, 0);
        }

        // Forked as a child
        if (pid == 0) {
                sleep(1);
                client("localhost", "8080");
        }

        // Something went wrong with forking
        if (pid < 0) {
                perror("fork");
        }
}

And the program execution is as simple as…

# clang net.c -o net
# ./net
client received: hello world

Basic C Programming on FreeBSD

Standard

In this article, I share some programming techniques with the C programming language on FreeBSD.  As of FreeBSD 11.0, the basic installation comes with clang(1) tools for programming in C, C++, and Obj-C languages.

When I was a teaching assistant, I used to give my students a crash course in C programming.  They already know C++ from their year 1.  Once they understand the differences, they picked up C real quick.

C is one of my favourite programming languages.  It’s features are minimal yet comprehensive.  Because of its simplicity, it is also very popular in system implementation.  Once somebody understands it, he is able to read and understand a lot of system implementation work.  (Warning: he will also start to dislike some particular kernel implementations, because they are really badly written.)

Why is programming related to this blog?  Indeed, one of the long-term goals here is to build a distributed software transactional memory system and deploy it on the cloud.  Time will tell if I am capable of this accomplishment.

As a side note, in Cantonese, “C” has similar pronunciation as “詩” (poem), which is a popular character for feminine names, such as “李慧詩“.  It is quite romantic to say “寫詩” (write poems) in place of “programming in that low-level language”… unless one broke up with one of those girls.  Ah.  The weather today is no good; and life is really harsh!  Let’s carry (not Carrie) on.

The Euclidean Algorithm

Today, we take the Euclidean Algorithm as an example.  It is useful for finding the greatest common divisor (GCD) of two integers.  I am not a mathematician, so please let me jump to my conclusion, as copied from the Wikipedia:

function gcd(a, b)
    while b ≠ 0
       t := b; 
       b := a mod b; 
       a := t; 
    return a;

In short:  There are two numbers.  The larger number is replaced with the modulus between the two number.  The logic repeats until a number turns into nothing.  The remaining number is the answer.

The First Attempt

Open up your favourite text editor and write to a file “gcd.c”.  I use vi(1), but you can always use the easier ee(1).  There are two functions in this program.  The “gcd” function is responsible computing the greatest common divisor.  The “main” function is the entry point.  It uses scanf(3) to input two integers, “a” and “b”, and then uses printf(3) to output the result.  In order to use these two functions, a header file “stdio.h” has to be included, as hinted by the manual pages.

There are “&” signs near the “a” and “b” and they mean pass-by-pointer.  Normally, variables are only pass-by-value.  The pass-by-pointer strategy allows the scanf(3) function to take the pointers and update the variables directly.  The arguments “argc” and “argv” are the argument counts (a number) and argument vectors (array of strings, or array of array of characters).  We do not use the arguments so we can leave them untouched.

#include <stdio.h>

int gcd(int a, int b) {
        int c;
        while (b != 0) {
                c = a % b;
                a = b;
        }
}

int main(int argc, char** argv) {
        int a, int b;
        scanf("%d %d", &a, &b);
        printf("%d\n", gcd(a, b));
        return 0;
}

Compilation Errors

Compile with the cc(1) command.  Like the way I demonstrated to my students, the first attempt is usually a failed one…

# cc gcd.c -o gcd
gcd.c:9:1: warning: control reaches end of non-void function
}
^
gcd.c:12:9: error: expected identifier or '('
        int a, int b;
               ^
gcd.c:12:8: error: expected ';' at end of declaration
        int a, int b;
              ^
              ;
1 warning and 2 errors generated.

The Second Attempt

It seems that semicolon should be used to separate declaration of the two variables.  (I know, comma works another way, but I am not going to tell…)

#include <stdio.h>

int gcd(int a, int b) {
        int c;
        while (b != 0) {
                c = a % b;
                a = b;
        }
}

int main(int argc, char** argv) {
        int a;  // <-- this line
        int b;  // <-- this line
        scanf("%d %d", &a, &b);
        printf("%d\n", gcd(a, b));
        return 0;
}

Runtime Error

Let’s compile and try again.  The program does compile despite the warning.  For dramatic sake, ignore the warning and try.  In the smoke test, we are supposed to enter the two numbers into the standard input.  After inputting the two numbers, press enter a few times and …

# cc gcd.c -o gcd
gcd.c:9:1: warning: control reaches end of non-void function
}
^
1 warning generated.
# ./gcd
4 24

the program loops forever, press Ctrl-C to break out.

First Time Debugging

To debug, let us use the LLVM debugger. (This is a new standard feature in FreeBSD 11.0.  We used to have another debugger in the past.)  Issue the lldb command, run with command “run”, input some text, then break it with Ctrl-C.  Unlike last time we break to the shell, this time we end up seeing some assembly code dump.

# lldb gcd
(lldb) target create "gcd"
Current executable set to 'gcd' (x86_64).
(lldb) run
Process 93041 launching
Process 93041 launched: '/root/gcd' (x86_64)
4 24

^C
Process 93041 stopped
* thread #1: tid = 100067, 0x00000000004007ab gcd`gcd + 27, stop reason = signal SIGSTOP
    frame #0: 0x00000000004007ab gcd`gcd + 27
gcd`gcd:
->  0x4007ab <+27>: movl   %edx, -0x10(%rbp)
    0x4007ae <+30>: movl   -0xc(%rbp), %edx
    0x4007b1 <+33>: movl   %edx, -0x8(%rbp)
    0x4007b4 <+36>: jmp    0x40079a                  ; <+10>
(lldb) quit
Quitting LLDB will kill one or more processes. Do you really want to proceed: [Y/n] y

As part of my plan, you see nothing useful unless you understand assembly code.  Quit and say yes to confirm.  Then, we try compiling again with debug symbols.

# cc gcd.c -o gcd
gcd.c:9:1: warning: control reaches end of non-void function
}
^
1 warning generated.
# lldb gcd
(lldb) target create "gcd"
Current executable set to 'gcd' (x86_64).
(lldb) run
Process 93075 launching
Process 93075 launched: '/root/gcd' (x86_64)
4 24

^C
Process 93075 stopped
* thread #1: tid = 100096, 0x00000000004007ab gcd`gcd(a=24, b=24) + 27 at gcd.c:6, stop reason = signal SIGSTOP
    frame #0: 0x00000000004007ab gcd`gcd(a=24, b=24) + 27 at gcd.c:6
   3   int gcd(int a, int b) {
   4     int c;
   5     while (b != 0) {
-> 6       c = a % b;
   7       a = b;
   8     }
   9   }
(lldb)

To step through the program, use the command “print” to print a variable content, and “next” to step to the next statement.  Repeat this a few times, we see the variable content does not change.  The condition to break out of the loop will not be satisfiable.

(lldb) print b
(int) $0 = 24
(lldb) next
Process 93075 stopped
* thread #1: tid = 100096, 0x00000000004007ae gcd`gcd(a=24, b=24) + 30 at gcd.c:7, stop reason = step over
    frame #0: 0x00000000004007ae gcd`gcd(a=24, b=24) + 30 at gcd.c:7
   4     int c;
   5     while (b != 0) {
   6       c = a % b;
-> 7       a = b;
   8     }
   9   }
   10  
(lldb) print b
(int) $1 = 24
(lldb) next
Process 93075 stopped
* thread #1: tid = 100096, 0x00000000004007b4 gcd`gcd(a=24, b=24) + 36 at gcd.c:5, stop reason = step over
    frame #0: 0x00000000004007b4 gcd`gcd(a=24, b=24) + 36 at gcd.c:5
   2   
   3   int gcd(int a, int b) {
   4     int c;
-> 5     while (b != 0) {
   6       c = a % b;
   7       a = b;
   8     }
(lldb) print b
(int) $2 = 24
(lldb) next
Process 93075 stopped
* thread #1: tid = 100096, 0x00000000004007a4 gcd`gcd(a=24, b=24) + 20 at gcd.c:6, stop reason = step over
    frame #0: 0x00000000004007a4 gcd`gcd(a=24, b=24) + 20 at gcd.c:6
   3   int gcd(int a, int b) {
   4     int c;
   5     while (b != 0) {
-> 6       c = a % b;
   7       a = b;
   8     }
   9   }
(lldb) print b
(int) $3 = 24

The Third Attempt

In order to make the loop terminate, the variable “b” must change.  We take revision on the source code and noticed a line is missing.  Here is another iteration:

#include <stdio.h>

int gcd(int a, int b) {
        int c;
        while (b != 0) {
                c = a % b;
                a = b;
                b = c;  // <-- this line
        }
}

int main(int argc, char** argv) {
        int a;
        int b;
        scanf("%d %d", &a, &b);
        printf("%d\n", gcd(a, b));
        return 0;
}

Second Time Debugging

I am going to save time and tell you the code does not work.  It consistently returns “0” in my case.  Let us jump into the debugger again.  Because the program does not loop indefinitely this time, we need another way to stop the program before it finishes.  In this example, I used a “breakpoint set” command with “–file” and “–line” option.  Then I used “step” command to step inside the “gcd” function call.

# cc gcd.c -o gcd
gcd.c:9:1: warning: control reaches end of non-void function
}
^
1 warning generated.
# lldb gcd
(lldb) target create "gcd"
Current executable set to 'gcd' (x86_64).
(lldb) breakpoint set --file gcd.c --line 15
Breakpoint 1: where = gcd`main + 53 at gcd.c:15, address = 0x0000000000400805
(lldb) run
Process 93155 launching
Process 93155 launched: '/root/gcd' (x86_64)
4 24

^C
Process 93155 stopped
* thread #1: tid = 100085, 0x0000000000400805 gcd`main(argc=1, argv=0x00007fffffffeb60) + 53 at gcd.c:15, stop reason = breakpoint 1.1
    frame #0: 0x0000000000400805 gcd`main(argc=1, argv=0x00007fffffffeb60) + 53 at gcd.c:15
   12  int main(int argc, char** argv) {
   13  int a; int b;
   14  scanf("%d %d", &a, &b);
-> 15  printf("%d\n", gcd(a, b));
   16  return 0;
   17  }
(lldb) print a
(int) $0 = 4
(lldb) print b
(int) $1 = 24
(lldb) step
Process 93155 stopped
* thread #1: tid = 100085, 0x000000000040079a gcd`gcd(a=4, b=24) + 10 at gcd.c:5, stop reason = step in
    frame #0: 0x000000000040079a gcd`gcd(a=4, b=24) + 10 at gcd.c:5
   2   
   3   int gcd(int a, int b) {
   4     int c;
-> 5     while (b != 0) {
   6       c = a % b;
   7       a = b;
   8       b = c;
(lldb) next
Process 93155 stopped
* thread #1: tid = 100085, 0x00000000004007a4 gcd`gcd(a=4, b=24) + 20 at gcd.c:6, stop reason = step in
    frame #0: 0x00000000004007a4 gcd`gcd(a=4, b=24) + 20 at gcd.c:6
   3   int gcd(int a, int b) {
   4     int c;
   5     while (b != 0) {
-> 6       c = a % b;
   7       a = b;
   8       b = c;
   9     }
(lldb) next
Process 93155 stopped
* thread #1: tid = 100085, 0x00000000004007ae gcd`gcd(a=4, b=24) + 30 at gcd.c:7, stop reason = step in
    frame #0: 0x00000000004007ae gcd`gcd(a=4, b=24) + 30 at gcd.c:7
   4     int c;
   5     while (b != 0) {
   6       c = a % b;
-> 7       a = b;
   8       b = c;
   9     }
   10  }
(lldb) next
Process 93155 stopped
* thread #1: tid = 100085, 0x00000000004007b4 gcd`gcd(a=24, b=24) + 36 at gcd.c:8, stop reason = step over
    frame #0: 0x00000000004007b4 gcd`gcd(a=24, b=24) + 36 at gcd.c:8
   5     while (b != 0) {
   6       c = a % b;
   7       a = b;
-> 8       b = c;
   9     }
   10  }
   11  
(lldb) next
Process 93155 stopped
* thread #1: tid = 100085, 0x00000000004007ba gcd`gcd(a=24, b=4) + 42 at gcd.c:5, stop reason = step over
    frame #0: 0x00000000004007ba gcd`gcd(a=24, b=4) + 42 at gcd.c:5
   2   
   3   int gcd(int a, int b) {
   4     int c;
-> 5     while (b != 0) {
   6     c = a % b;
   7     a = b;
   8     b = c;
(lldb) next
Process 93155 stopped
* thread #1: tid = 100085, 0x00000000004007a4 gcd`gcd(a=24, b=4) + 20 at gcd.c:6, stop reason = step over
    frame #0: 0x00000000004007a4 gcd`gcd(a=24, b=4) + 20 at gcd.c:6
   3   int gcd(int a, int b) {
   4     int c;
   5     while (b != 0) {
-> 6       c = a % b;
   7       a = b;
   8       b = c;
   9     }
(lldb) next
Process 93155 stopped
* thread #1: tid = 100085, 0x00000000004007ae gcd`gcd(a=24, b=4) + 30 at gcd.c:7, stop reason = step over
    frame #0: 0x00000000004007ae gcd`gcd(a=24, b=4) + 30 at gcd.c:7
   4     int c;
   5     while (b != 0) {
   6       c = a % b;
-> 7       a = b;
   8       b = c;
   9     }
   10  }
(lldb) next
Process 93155 stopped
* thread #1: tid = 100085, 0x00000000004007b4 gcd`gcd(a=4, b=4) + 36 at gcd.c:8, stop reason = step over
    frame #0: 0x00000000004007b4 gcd`gcd(a=4, b=4) + 36 at gcd.c:8
   5     while (b != 0) {
   6       c = a % b;
   7       a = b;
-> 8       b = c;
   9     }
   10  }
   11  
(lldb) next
Process 93155 stopped
* thread #1: tid = 100085, 0x00000000004007ba gcd`gcd(a=4, b=0) + 42 at gcd.c:5, stop reason = step over
    frame #0: 0x00000000004007ba gcd`gcd(a=4, b=0) + 42 at gcd.c:5
   2   
   3   int gcd(int a, int b) {
   4     int c;
-> 5     while (b != 0) {
   6       c = a % b;
   7       a = b;
   8       b = c;
(lldb) next
Process 93155 stopped
* thread #1: tid = 100085, 0x00000000004007bf gcd`gcd(a=4, b=0) + 47 at gcd.c:10, stop reason = step over
    frame #0: 0x00000000004007bf gcd`gcd(a=4, b=0) + 47 at gcd.c:10
   7       a = b;
   8       b = c;
   9     }
-> 10  }
   11  
   12  int main(int argc, char** argv) {
   13  int a; int b;
(lldb) next
Process 93155 stopped
* thread #1: tid = 100085, 0x0000000000400813 gcd`main(argc=1, argv=0x00007fffffffeb60) + 67 at gcd.c:15, stop reason = step over
    frame #0: 0x0000000000400813 gcd`main(argc=1, argv=0x00007fffffffeb60) + 67 at gcd.c:15
   12  int main(int argc, char** argv) {
   13    int a; int b;
   14    scanf("%d %d", &a, &b);
-> 15    printf("%d\n", gcd(a, b));
   16    return 0;
   17  }

We see the values of variables “a” and “b” decreases alone the time line.  Pay attention to the transition between the “gcd” function to the “main” function.  There is not a return value.  This is why we kept having the warning.  Compiler warnings are often more useful than what programmers think.  Sometimes, I am amazed why the open source software packages contain some many warnings in compilations and they still work.

The Final Version

We get to return the answer in the “gcd” function.  Here is how it the code being corrected.

#include <stdio.h>

int gcd(int a, int b) {
        int c;
        while (b != 0) {
                c = a % b;
                a = b;
                b = c;
        }
        return a;  // <-- this line
}

int main(int argc, char** argv) {
        int a;
        int b;
        scanf("%d %d", &a, &b);
        printf("%d\n", gcd(a, b));
        return 0;
}

The Final Testing

# cc gcd.c -o gcd
# ./gcd
0 5
5
# ./gcd
5 0
5
# ./gcd
1 9
1
# ./gcd
9 1
1
# ./gcd
4 36
4
# ./gcd
36 4
4
# ./gcd
24 36
12
# ./gcd
36 24
12